Squeezing for Advanced LIGO L Barsotti LIGO Laboratory

Squeezing for Advanced LIGO L. Barsotti LIGO Laboratory – MIT G 1500734 Enhanced Interferometers Session --- May 18, 2015

Lessons learned from GEO and LIGO ² Losses are very unforgiving: they come in many forms, some of them are very hard to fix: OPO cavity, Faradays, mode mismatch, OMC throughput, photodiode quantum efficiency, misalignments, … ² When coupling a squeezed light source to an interferometer, the ``usual” nasty noise couplings show up: lock error point, ``phase noise” back scattered noise, … ² You needs some cleverness in your control scheme GW Signal Dwyer et al. , Optics Express, Vol. 21, Issue 16, pp. 19047 -19060 (2013) Grote et al. , Phys. Rev. Lett. 110, 181101 (2013) Chua et al. , Class. Quantum Grav. 31 035017 (2014) Dooley et al. , Optics Express Vol. 23, Issue 7, pp. 8235 -8245 (2015) Schreiber et al. , LIGO-P 1500056 (2015) 2

Squeezing R&D program 1. Study filter cavity performance and demonstrate frequency dependent squeezing in the audio-frequency region 2. Incorporate lessons learned from GEO/LIGO in a squeezed light source targeted for a. LIGO 3. Couple squeezing + filter cavity in a design compatible with an early upgrade to a. LIGO

Study of filter cavity performance ² Direct measurements of losses in a high finesse cavity : * Isogai et. al. Optics Express 21 (2013) * Optical scatter of quantum noise filter cavity optics (J. Smith’s group @ Fullerton, http: //arxiv. org/pdf/1411. 5403 v 1. pdf) Experimental results (~10 ppm round trip losses for 1 -3 mm beam size) compatible with a “short” filter cavity for Advanced LIGO ² Development of model to describe filter cavity noise mechanisms, including loss and mode mismatch P. Kwee, J. Miller, et al. Phys. Rev. D 90, 062006 A “short” filter cavity allows squeezing injection without degrading low frequency sensitivity (“not harm”) (see Tomoki’s talk tomorrow) 5

Demonstration of frequency dependent squeezing with a 2 m filter cavity @ MIT (2015) Paper circulated to the LSC: P 1500062 Extrapolation for a. LIGO 16 m filter cavity: factor of 2 reduction in shot noise (6 d. B), 25% reduction in radiation pressure noise (2 d. B) (see Tomoki’s talk tomorrow) 6

Frequency Dependent Squeezing (“short” filter cavity) High frequency improvement, + 25% BNS-BNS range (200 vs 250 Mpc) Enables further improvement through coating thermal noise reduction 7

Squeezing R&D program 1. Study filter cavity performance and demonstrate frequency dependent squeezing in the audio-frequency region 2. Incorporate lessons learned from GEO/LIGO in a squeezed light source targeted for a. LIGO 3. Couple squeezing + filter cavity in a design compatible with an early upgrade to a. LIGO

Conceptual design of a new squeezed light source for Advanced LIGO ² OPO typically sits on an in-air table, no seismic isolation, no acoustic enclosure ² Advantages of seismic and acoustic isolation: reduce OPO cavity length noise (thus phase noise), mitigate impact from back scattering, less stringent requirements on alignment stability, OPO can be used as “quiet reference cavity” for the pump laser Let’s move the OPO into the vacuum envelope on seismic isolated tables E. Oelker et al. , Optics Express, Vol. 22, Issue 17, pp. 21106 -21121 (2014)

A new squeezed light source (ANU/MIT) ² First monolithic Optical Parametric Oscillator built at ANU ² Same optical characteristics as OPO used in H 1 Squeezing test (bow-tie, doubly resonant) ANU OPO cavity More than 8 d. B of squeezing measured in-vacuum (see Georgia Mansell’s talk at the last LVC: G 1500364) Observation that dual resonance condition shifts air/in-vacuum (dispersion problem: air refractive index change with wavelength)

A new squeezed light source (ANU/MIT) ²New OPO just built at MIT, incorporating lessons learned from first ANU prototype (in particular, remotely controllable translation stage for the OPO crystal to compensate dispersion problem) ²Option of fiber coupling light to the OPO under investigation ü New control scheme and noise budget analysis in progress ü Results from table top experiment at MIT expected by the end of the summer ü Investigations on-going in parallel at ANU New OPO cavity @ MIT Picture Credit: Georgia Mansell

Squeezing R&D program 1. Study filter cavity performance and demonstrate frequency dependent squeezing in the audio-frequency region 2. Incorporate lessons learned from GEO/LIGO in a squeezed light source targeted for a. LIGO 3. Couple squeezing + filter cavity in a design compatible with an early upgrade to a. LIGO

Still a “Conceptual Design” Basic idea for implementation in a. LIGO, include the option for a filter cavity

One of the greatest challenge: reduce optical loss ² Easier said than done ² On-going LSC “low loss” effort T 1400715 ² https: //wiki. ligo. org/AIC/Low. Loss (GEO’s talk earlier, Kate/Emil’s talk tomorrow)

Plausible Scenario for Squeezing in a. LIGO ²Modest squeezing performance with a. LIGO not yet at full power (maybe post O 2 (2017? ), maybe without filter cavity) ²Incorporate filter cavity and low loss readout (maybe post O 3 (2018? )) No decision yet, detector performance & network observing runs set the schedule

The Global Perspective Possible Upgrade to Advanced LIGO 16

The Global Perspective 17

The Message ² Squeezing R&D progressing well: ² First demonstration of audio-band frequency dependent squeezing successful, no surprises ² ANU/MIT effort on new squeezed light source on-going ² LSC effort to attack loss ² We believe this is the way to go…proposal for a full scale squeezing injection system at MIT in the near future ² Effort will ramp up soon to converge on preliminary design for a. LIGO ² Plan for actual implementation in a. LIGO being developed (plausible scenario: post-O 2), detector performance & network observing runs set the schedule More in the Topologies & Squeezing workshop. .

Summary of Squeezing Options Option Benefit & Cost Readiness Frequency Independent Squeezing x 2 improvement at HF, worse low frequency system development $1 M / IFO Frequency Dependent Squeezing (short cavity) x 2 improvement at HF, preserve low frequency Frequency Dependent Squeezing (long cavity) x 2 improvement at HF, improvement at low frequency too add $1 M / IFO (TBC) system development add $500 k / IFO technology development 19

Extra Slides 20

Readiness level / cost for Squeezing ² Frequency independent ü Already applied in large scale interferometers Nature Physics 7, 962 (2011), Nature Photonics 7, 613– 619 (2013) ü Mature technology: system development phase ü High frequency improvement, risk mitigation for high power operation in a. LIGO ü Tentative cost estimate: $1 M per interferometer ² Frequency dependent (“short cavity”) Recent demonstration with table top experiment (P 1500062) Mature technology: system development phase +25% improvement in BNS-BNS range (~250 Mpc) Greater benefit when combined with reduced coating thermal noise (see Stefan’s talk, and Phys. Rev. D 91, 062005) ü Tentative estimate: additional $0. 5 M per interferometer ü ü ² Frequency dependent (“long cavity”) ü Particular beneficial for low frequency sources, when combined with other noise improvements ü Technology development phase; more costly 21

Frequency Dependent Squeezing - I GW Signal RADIATION PRESSURE NOISE ~30 Hz Quantum Noise SHOT NOISE High finesse detuned “filter cavity” which rotates the squeezing angle as function of 22 frequency

Long vs Short filter cavity (Nothing comes cheap) ²Advanced LIGO needs a a filter cavity with 50 Hz bandwidth ²Losses in a filter cavity deteriorate, if too high, make the filter cavity useless… Per-round-trip loss depends on the beam spot size (big beam size higher scatter losses), which depends on L 1 ppm/m

Focus on High Frequency Sources NS EOS LMXB SN Parameter estimation of compact binary systems: Phys. Rev. D 91, 044032 L. Barsotti, LIGO Laboratory - MIT 24

Limiting noise: quantum noise 25

Quantum shot noise limits the high frequency sensitivity RADIATION PRESSURE NOISE P = stored power m = mirror mass SHOT NOISE RADIATION PRESSURE NOISE: Back-action noise caused by random motion of optics due to fluctuations of the number of impinging photons Additional displacement noise SHOT NOISE: Photon counting noise due to fluctuations of the number of photon detected at the interferometer output Limitation of the precision to measure 26 arm displacement:

Reducing losses is critical for achieving 6 - 10 d. B (x 2 -3 shot noise reduction) LIGO GEO We need less than 20% total losses a. LIGO readout now has ~30%(w/o squeezing) GW Signal 27

Frequency Independent Squeezing RADIATION PRESSURE NOISE gets worse SHOT NOISE gets better by a factor of 2 High frequency improvement, no benefit in BNS-BNS range 28

Frequency Dependent Squeezing (“short” filter cavity) High frequency improvement, + 25% BNS-BNS range (200 vs 250 Mpc) Enables further improvement through coating thermal noise reduction 29

Frequency Dependent Squeezing (“long”, or low loss, filter cavity) More challenging than “short cavity”; particularly beneficial for targeting low/mid frequency sources, especially when combined with other 30

Signal Recycling Detuning ² In principle, ability to target high frequency sources without squeezing ² Less hardware investment with respect to squeezing, but challenge from the controllability of the interferometer ² Given the same loss in the interferometer, benefit at high frequency is comparable to frequency dependent squeezing in a narrow band, worse elsewhere Signal recycling detuning not particular beneficial for high frequency sources Interesting cases for low-mid frequencies regions, especially when combined with frequency independent squeezing 31

Balanced Homodyne Detection ²Standard technique in table top squeezing experiments ²It has advantages compared to DC readout when applied to large scale interferometers Optics Express Vol. 22, Issue 4, pp. 4224 -4234 (2014) ²Main advantage: remove static carrier field at the antisymmetric port 32

Balanced Homodyne Detection L 1 current high frequency noise budget Optics Express Vol. 22, Issue 4, pp. 4224 -4234 (2014) Credit: Denis Martynov 33

LOSS ²Faraday Isolators: need to be < 1% loss per pass a. LIGO output faraday now adds 4% loss per pass ²OMC loss: ideally less than 1% study done by Koji Arai @ Caltech a. LIGO OMC loss from 3% – 7%, it strongly depends on beam spot position on the optics ²Mode matching: target is 1%-2% the sad reality is that mode matching is hard (best ever LIGO mode matching to the OMC: 5% loss, L 1 e. LIGO) ²Photodiode quantum efficiency = 99% ? ? Hartmut Grote’s wisdom: “only 1 batch of diodes have ever been measured with only 1% loss @ 1064” 34

Signal Recycling Detuning with frequency independent squeezing 35

Signal Recycling Detuning with frequency independent squeezing, low loss 36